How FoxBetz Works
FoxBetz scans Ekstraklasa matches daily at 9:00 AM and finds value bets — situations where our model's probability is higher than what the bookmaker's odds imply. When the edge exceeds 7%, you get an alert.
Step 1: Set Your Bankroll
Your bankroll is the total amount of money you've set aside exclusively for betting. This is money you can afford to lose completely.
Common starting bankrolls: 100 USD, 500 USD, 1000 USD. Pick an amount, write it down, and stick with it.
Step 2: Understanding the Alert
When FoxBetz finds a value bet, you'll see something like this:
The key number is the Suggested Stake %. This tells you how much of your current bankroll to bet.
Step 3: Calculate Your Stake
Multiply the suggested percentage by your current bankroll (not your starting amount).
Example: Starting bankroll 500 USD
| Bet # | Current Bankroll | Bot Says | Your Stake | Result | New Bankroll |
|---|---|---|---|---|---|
| 1 | 500.00 USD | 3.5% @ 2.12 | 17.50 USD | WIN +19.60 USD | 519.60 USD |
| 2 | 519.60 USD | 6.4% @ 2.80 | 33.25 USD | LOSS -33.25 USD | 486.35 USD |
| 3 | 486.35 USD | 2.8% @ 1.98 | 13.62 USD | WIN +13.35 USD | 499.70 USD |
Quick Reference: Stake per Bankroll Size
Don't want to do math? Here's a cheat sheet for common bankroll sizes.
| Bot Says | 100 USD | 250 USD | 500 USD | 1000 USD | 2000 USD |
|---|---|---|---|---|---|
| 1.5% | 1.50 | 3.75 | 7.50 | 15.00 | 30.00 |
| 2.0% | 2.00 | 5.00 | 10.00 | 20.00 | 40.00 |
| 3.0% | 3.00 | 7.50 | 15.00 | 30.00 | 60.00 |
| 4.0% | 4.00 | 10.00 | 20.00 | 40.00 | 80.00 |
| 5.0% | 5.00 | 12.50 | 25.00 | 50.00 | 100.00 |
| 6.0% | 6.00 | 15.00 | 30.00 | 60.00 | 120.00 |
| 8.0% | 8.00 | 20.00 | 40.00 | 80.00 | 160.00 |
Values in USD. Remember: use your CURRENT bankroll, not your starting amount.
Multiple Bets on the Same Day
If the bot finds 3 value bets on the same matchday:
Example: bankroll is 500 USD, bot finds 3 bets at 3%, 2%, and 4%.
- Bet A: 500 x 3% = 15.00 USD
- Bet B: 500 x 2% = 10.00 USD
- Bet C: 500 x 4% = 20.00 USD
After all three settle, update your bankroll with the combined result. Use the new number for the next day's bets.
Flat Staking (Alternative)
If Quarter-Kelly feels complicated, you can use flat staking instead:
- Conservative: Bet 2% of your starting bankroll on every alert (e.g. 10 USD on a 500 USD bankroll)
- Aggressive: Bet 4% of your starting bankroll (e.g. 20 USD on a 500 USD bankroll)
Flat staking is simpler but doesn't compound as well. The bot's Quarter-Kelly suggestion is the recommended approach for long-term growth.
Where to Place Bets
FoxBetz tells you which bookmaker has the best odds. The alert shows something like "Best Odds: 2.27 (1xBet)".
Licensed Polish bookmakers:
- STS (sts.pl)
- Fortuna (efortuna.pl)
- Betclic (betclic.pl)
- Superbet (superbet.pl)
- Betfan (betfan.pl)
- LV BET (lvbet.pl)
Timing
- Alerts go out at 9:00 AM daily.
- Place your bets as soon as you see the alert. Odds move throughout the day — early = better odds.
- Never bet after kickoff (obviously).
- If the odds have dropped significantly since the alert, the edge may no longer exist. Skip it.
Common Mistakes
Chasing losses
After a losing streak, resist the urge to increase stakes. Quarter-Kelly automatically reduces your stake as your bankroll shrinks — trust the system.
Skipping "boring" bets
An Under 2.5 bet at 1.95 odds might not be exciting, but if the model says there's a 12% edge, it's a better bet than a flashy 5.00 odds bet with 3% edge. The edge is what matters, not the payout.
Betting with money you can't afford to lose
Even with a +18% yield model, you WILL have losing streaks. The backtest shows drawdowns of 15-20% are normal. If a 100 USD loss would stress you out, your bankroll is too big.
Not tracking results
FoxBetz tracks everything automatically. Check the All Bets page and Case Study regularly. If the model stops performing, you'll see it in the numbers.
About the Model
FoxBetz uses an ensemble model combining:
- Poisson Model (80%) — estimates expected goals per team using xG (expected goals) data with exponential time decay. Computes scoreline probabilities for every possible result (0-0 through 6-6).
- XGBoost (20%) — gradient-boosted ML model using rolling form, head-to-head records, days of rest, and shot statistics. Calibrated with isotonic regression.
The model is retrained daily on the latest data. xG data refreshes weekly from an open-source dataset.
Past performance does not guarantee future results. Bet responsibly. 18+.